Try each one before opening the answer. Every option is explained, with the Google Cloud documentation page that proves it.
Question 1 · choose 1
A Public NAT gateway with static port allocation serves a subnet of 200 GKE nodes, and it translates both the subnet's primary range and the Pod secondary range, which the nodes use as alias IP ranges larger than /32. The minimum ports per VM is set to 256. Every node must get its full reservation, and the partner allow-lists each NAT IP address in advance. How many NAT IP addresses are needed?
A1
B7
C4
D3
Show the answer and why
A1
Incorrect
One address would suffice at 256 ports per node, but with alias IP ranges Cloud NAT raises the minimum to 1,024.
B7
Incorrect
Seven assumes each node gets twice its minimum, which is how Private NAT allocates, not Public NAT.
C4
Correct
With the minimum raised to 1,024, each address's 64,512 ports cover 63 nodes, so 200 nodes need four addresses.
D3
Incorrect
Three addresses at 63 nodes each cover only 189 nodes, leaving 11 without their reservation.
Each NAT IP address offers 64,512 TCP and 64,512 UDP source ports. With static port allocation, when the gateway translates alias IP ranges larger than /32, Cloud NAT uses the larger of the configured minimum and 1,024 as the minimum ports per VM. 64,512 / 1,024 gives 63 VMs per address, and 200 / 63 rounds up to 4.
A Public NAT gateway uses manually assigned NAT IP addresses. One address must be retired, but long-running downloads that use it must not be cut off. What should the network engineer do with that address?
ADrain it, so the gateway stops using it for new connections but keeps it for established ones
BRemove it from the gateway at once; its established connections move to the other addresses
CSwitch the gateway to automatic NAT IP allocation, which migrates connections
DTurn off endpoint-independent mapping so connections can change address
Show the answer and why
ADrain it, so the gateway stops using it for new connections but keeps it for established ones
Correct
Draining stops new connections on the address while established connections close normally.
BRemove it from the gateway at once; its established connections move to the other addresses
Incorrect
Removing a manually assigned NAT IP address breaks its established connections immediately.
CSwitch the gateway to automatic NAT IP allocation, which migrates connections
Incorrect
Changing the allocation method isn't the documented way to retire one address gracefully.
DTurn off endpoint-independent mapping so connections can change address
Incorrect
Mapping behavior doesn't let an existing connection change its NAT address.
With manual NAT IP assignment you can either remove an address, which breaks its established connections immediately, or drain it: the gateway stops using the drained address for new connections but keeps using it for established ones, which can then close normally.
A Public NAT gateway is created with static port allocation, and nobody sets the minimum number of ports per VM. What minimum does Cloud NAT use for VMs without alias IP ranges?
A32 ports per VM
B64 ports per VM
C512 ports per VM
DAll ports of one NAT IP address per VM
Show the answer and why
A32 ports per VM
Incorrect
32 is the default for dynamic port allocation.
B64 ports per VM
Correct
Without a configured value, static port allocation uses a default minimum of 64 ports per VM.
C512 ports per VM
Incorrect
512 isn't a default for either allocation method.
DAll ports of one NAT IP address per VM
Incorrect
Ports are shared out per VM, not a whole address each.
If you don't specify a minimum number of ports per VM, Cloud NAT uses the default: 64 for static port allocation and 32 for dynamic port allocation. Public NAT uses static allocation by default; Private NAT uses dynamic.
A team deploys Secure Web Proxy in explicit proxy routing mode to control VMs' web access. What must happen on the VMs for their traffic to pass through the proxy?
ANothing: the VPC network's routes send all egress traffic to the proxy automatically
BA Cloud NAT gateway must be enabled on each VM
CEach VM's firewall must block port 443 so that traffic falls back to the proxy
DThe VMs or their applications must be configured to point to the proxy
Show the answer and why
ANothing: the VPC network's routes send all egress traffic to the proxy automatically
Incorrect
Automatic routing to the proxy describes next hop mode, not explicit proxy mode.
BA Cloud NAT gateway must be enabled on each VM
Incorrect
Cloud NAT isn't configured per VM and doesn't send traffic to the proxy.
CEach VM's firewall must block port 443 so that traffic falls back to the proxy
Incorrect
Blocking traffic doesn't make clients use a proxy.
DThe VMs or their applications must be configured to point to the proxy
Correct
In explicit proxy routing mode, workloads and clients are configured to point directly to the proxy server.
Secure Web Proxy has two routing modes. In explicit proxy routing mode, you configure workloads and clients to point directly to the proxy, which then isolates them from the internet. In next hop mode, network routing sends outbound traffic to the proxy without client configuration.